Selection of good camera positions has many applications in Computer Graphics. It can be used to compute a walkthrough inside a scene that shows a higher amount of information or to select a minimal set of views for Image- Based Rendering. However, the selection of a good view is costly, even when using OpenGL for fast rendering, as it is necessary to analyze a high amount of camera positions. In this paper we show how histograms, an OpenGL extension available in current graphics hardware, may be used together with an adaptive algorithm in order to obtain on the fly best views for objects of moderate complexity (several thousands of polygons).